The aim of this master’s thesis was to create a control system for laboratory distillation column model. There are two different models considered. Real and software model. For the real model suitable components were selected and the model was built. Software model allows safe testing of control algorithm functionality without the need of real model. The simple visualization that works with both models was developed. Both distillation column models - real and software - were tested as a last phase of work.

Yes / Batch reactive distillation is well known for improved conversion and separation of desired reaction products. However, for a number of reactions, the distillation can separate the reactants depending on their boiling points of them and thus not only reduces the benefit of the reactive distillation but also offers operational challenges for keeping the reactants together. Methyl lactate (ML) synthesis via the esterification of lactic acid (LA) with methanol in a reactive distillation falls into this category and perhaps that is why this process has not been explored in the past. The boiling points of the reactants (LA, methanol) are about 490 K and 337 K while those of the products (ML, water) are 417 K and 373 K respectively. Clearly in a conventional reactive distillation (batch or continuous) methanol will be separated from the LA and will reduce the conversion of LA to ML significantly. Here, first the limitations of the use of conventional batch distillation column (CBD) for the synthesis of ML is investigated in detail and a semi-batch reactive distillation (SBD) configuration is studied in detail where LA is the limiting reactant and methanol is continuously fed in excess in the reboiler allowing the reactants to be together for a longer period. However, this poses an operational challenge that the column has to be carefully controlled to avoid overflow of the reboiler at any time of the operation. In this work, the performance of SBD for the synthesis of ML is evaluated using model based optimization in which operational constraints are embedded. The results clearly demonstrate the viability of the system for the synthesis of ML.

Yes / The synthesis of ethyl benzoate (EtBZ) via esterification of benzoic acid (BeZ) with ethanol in a reactive distillation is challenging due to complex thermodynamic behaviour of the chemical reaction and the difficulty of keeping the reactants together in the reaction zone (ethanol having the lowest boiling point can separate from the BeZ as the distillation proceeds) causing a significant decrease in the conversion of BeZ in a conventional reactive distillation column (batch or continuous). This might be the reason of not reporting the use of reactive distillation for EtBZ synthesis although the study of BeZ esterification reaction is available in the public literature. Our recently developed Integrated Conventional Batch Distillation (i-CBD) column offers the prospect of revisiting such reactions for the synthesis of EtBZ, which is the focus of this work. Clearly, i-CBD column outperforms the Conventional Batch Distillation (CBD) column in terms of product amount, purity and conversion of BeZ and eliminates the requirement of excess use of ethanol. For example, compared with CBD column, the i-CBD operation can yield EtBZ at a much higher purity (0.925 compared to 0.730) and can convert more benzoic acid (93.57% as opposed to only 74.38%).

Increased interest in unconventional batch distillation column configurations offers new opportunities for increasing the flexibility and energy efficiency of batch distillation. One configuration of particular interest is multivessel batch distillation column, which can be viewed as a generalization of all previously studied batch column configuration. In this work, for the first time the optimal design and operation tasks are developed for multivessel batch distillation with strict product specifications under fixed product demand. Also, in this work, two different operation schemes defined as STN (State Task Network) in terms of the option and numbers of off-cuts were considered for binary and ternary separation. Both the vapour load and number of stages in each column section together with the production sequence are optimised to achieve maximum profit function. The performance of the multivessel batch distillation column is evaluated against the performance of conventional batch column with a simple dynamic model using binary and ternary mixtures. It has been found that profitability improves with the multivessel system in both separations. gPROMS, a user-friendly, software is used for the modeling, simulation, and optimisation.

Recent work in dataset distillation focuses on distilling supervised classification datasets into smaller, synthetic supervised datasets in order to reduce per-model costs of training, to provide interpretability, and to anonymize data. Distillation and its benefits can be extended to a wider array of tasks. We propose a generalization of dataset distillation, which we call task distillation. Using techniques similar to those used in dataset distillation, any learning task can be distilled into a compressed synthetic task. Task distillation allows for transmodal distillations, where a task of one modality is distilled into a synthetic task of another modality, allowing a more complex learning task, such as a reinforcement learning environment, to be reduced to a simpler learning task, such as supervised classification. In order to advance task distillation beyond supervised-to-supervised distillation, we explore distilling reinforcement learning environments into supervised learning datasets. We propose a new distillation algorithm that allows PPO to be used to distill a reinforcement learning environment. We demonstrate k-shot learning on distilled cart-pole to demonstrate the effectiveness of our distillation algorithm, as well as to explore distillation generalization. We distill multi-dimensional cart-pole environments to their minimum-sized distillations and show that this matches the theoretical minimum number of data instances required to teach each task. We demonstrate how a distilled task can be used as an interpretability artifact, as it compactly represents everything needed to learn the task. We demonstrate the feasibility of distillation in more complex Atari environments by fully distilling Centipede and demonstrating that distillation is cheaper than training directly on Centipede for training more than 9 models. We provide a method to "partially" distill more complex environments and demonstrate it on Ms. Pac-Man, Pong, and Space Invaders and show how it scales distillation difficulty fully on Centipede.

Organic esters are an important class of industrial and commercial chemicals that can be found in solvents, plasticizer, food flavours, detergents, agrochemicals, and pharmaceuticals. The most common way to synthesis organic esters is with esterification or transesterification. Where esterification was the chosen method in this thesis.  This thesis provides a techno-economic assessment on the production of propyl acetate and glycerol trivalerate through different continuous routes that is than compared with batch production under the same conditions. Simulations was done on trivalerate due to limited literature and data on pentaerythritol tetravalerate. Different continuous technologies that have been assessed in this thesis was plug flow reactor (PFR), Reactive Distillation (RD), and Reactive – Extractive Distillation (RED). The production of mono- and polyol esters with different unit operators was simulated in Aspen Plus V.14. Techno-Economic analysis was conducted with APEA (Aspen Process Economic Analyzer), where cost of raw materials, products, and utilities was inserted to evaluate annual operating cost and product sale. Reaction kinetic for esterification of trivalerate was estimated by obtained values from simulations of a Gibbs reactor in Aspen Plus at four different temperatures. Kinetics was estimated by applying the relation between the chemical equilibrium constant and the Arrhenius equation. Where rate constant and activation energy for forward and revers reaction was obtained by varying min and max values for lsqcurvefit in MATLAB and then validate results with published kinetics.  Results from production of propyl acetate with batch, PFR and RED provides an annual profit of approximately 1 M$ at a capacity of 41.65 kton/year. The three different process provides also approximately an equal capital cost, operating cost, equipment cost, and installed cost according to APEA. RED provided the lowest propyl acetate yield at 93%, batch and PFR provided a propyl acetate yield of 94%.  Results from production of trivalerate with batch, PFR, and RD provides an annual profit of 5.4, 5.78, and 9.2 M$ at a capacity of approximately 5 kton/year. Where RD process provides the lowest capital cost, operating cost, equipment cost, and installed cost compared to batch and PFR processes according to APEA. Obtained results from production of trivalerate can be used to evaluate the economic and technical feasibility of a continuous production plant for pentaerythritol tetravalerate (PETV). Where initial simulations show a good economic and technical viability of a continuous ester production plant.

The genus Lippia Houst. (family Verbenaceae Juss.) has been used medicinally for centuries in various cultures. Numerous traditional medicinal applications of the South African Lippias are known, and among these is Lippia scaberrima Sond., the "Beukesbossie". Products recovered from L. scaberrima include essential oils by means of batch microwave distillation and a herbal tea manufactured by various methods. Microwave heating occurs through direct absorbtion of input energy. The whole sample is heated simultaneously with a heating rate much faster than for water distillation. Microwave distillation times can be expected to be less than corresponding water distillation times. A large number of parameters influence essential oil composition and yield. Various geographical, climatic, botanical, production and other parameters were identified and listed. The natural distribution of airdried leaves and flower heads and water as solvent was used in a domestic type microwave oven. Microwave distillation production parameters were studied and optima determined with respect to maximum oil yield. The batch microwave distillation production parameters studied (irradiation time, ratio of water used to plant material used, plant material moisture content, solvent recycling, physical size of plant material, addition of a surface tension modifier, soaking as pretreatment, load size, distillation kinetics and alternative solution medium namely ethanol) lead to a better understanding of microwave essential oil production. It enabled design of prototype continuous microwave distillation apparatus. Mobile, continuous microwave distillation equipment enables extraction of essential oils in a rural environment. It allows rapid investigation of a large variety of plants, and retrieval of essential oils in an economical and uncomplicated manner. Water distillation represents a classical technique for essential oil retrieval whereas microwave distillation is probably the most modern method. Physical and chemical properties of microwave and water distilled oils were determined. This enabled comparison of the microwave technique to a standard, industrially accepted production method. Tea was manufactured from L. scaberrima by means of the green tea, Rooibos tea and black tea methods as a first iteration in the production of a commercial tea. Drying curves was developed as well as appropriate regions of production variables like withering- and fermentation temperatures and times. / Dissertation (PhD(Ing.Chem))--University of Pretoria, 2002. / Chemical Engineering / unrestricted

Call number: LD2668 .T4 1985 C48 / Master of Science

Thesis (PhD)--University of Stellenbosch, 2004. / ENGLISH ABSTRACT: Structured packing is a popular column internal for both distillation and absorption unit operations. This is due to the excellent mass transfer characteristics and low pressure drop that it offers compared to random packing or trays. The main disadvantage is the lack in reliable models to describe the mass transfer characteristics of this type of packing. The recent development of the non-equilibrium model or rate based modelling approach has also emphasized the need for accurate hydraulic and efficiency models for sheet metal structured packing. The main focus of this study was to develop an accurate model for the mass transfer efficiency of Flexipac 350Y using a number of experimental and modelling techniques. Efficiency is however closely related to hydraulic capacity. Before attempting to measure and model the efficiency of Flexipac 350Y, the ability of existing published models to accurately describe the hydraulic capacity of this packing was tested. Holdup and pressure drop were measured using air/water and air/heavy paraffin as test systems. All experiments were performed on pilot plant scale 200mm ID glass columns. Satisfactory results were obtained with most of the models for determining the loading point and pressure drop for the air/water test system. All of the models tested predicted a conservative dependency of capacity on liquid viscosity for the air/paraffin test system. Efficiency and pressure drop were measured using the chlorobenzene/ethylbenzene test systems under conditions of total reflux in a 200mm ID glass column. Widely differing results were however obtained with the different models for the efficiency of Flexipac 350Y. Experiments were subsequently designed and performed to measure and correlate the vapour phase mass transfer coefficient and the effective surface area of Flexipac 350Y independently. The vapour phase mass transfer coefficient was measured and correlated by subliming naphthalene into air from coatings applied to specially fabricated 350Y gauze structured packing. The use of computational fluid dynamics (CFD) to model the vapour phase mass transfer coefficient is also demonstrated. The effective surface area for vapour phase mass transfer was measured with the chemical technique. The specific absorption rate of CO2 into monoethanolamine (MEA) using n-propanol as solvent was determined in a wetted-wall column and used to determine the effective surface area of Flexipac 350Y on pilot plant scale (200mm ID glass column). The efficiency of Flexipac 350Y could be modelled within an accuracy of 9% when using the correlations developed in this study and ignoringliquid phase resistance to mass transfer for the chlorobenzene/ethylbenzene test system under conditions of total reflux. The capacity and efficiency of the new generation high capacity packing Flexipac 350Y HC was also measured and compared with that of the normal capacity packing Flexipac 350Y. An increase in capacity of 20% was observed for the HC packing for the air/water system and 4% for the air/heavy paraffin system compared with the normal packing. For the binary total reflux distillation the increase in capacity varied between 8% and 15% depending on the column pressure. The gain in capacity was at the expense of a loss in efficiency of around 3% in the preloading region. / AFRIKAANSE OPSOMMING: Gestruktureerde pakking is 'n populêre pakkingsmateriaal en word algemeen gebruik in distillasie en absorpsie kolomme. Dit is hoofsaaklik as gevolg van die goeie massa-oordragseienskappe en lae drukval wat dit bied in vergelyking met 'random' pakking en plate. The hoof nadeel is egter die tekort aan akkurate modelle om die massa-oordrags eienskappe te bepaal. Om modelle te kan gebruik waar die massaoordragstempo direk gebruik word om gepakte hoogte te bepaal, word akkurate kapasiteits- en effektiwiteitsmodelle vir gestruktureerde plaatmetaalpakking benodig. Die hoof doelwit van hierdie studie was om 'n akkurate model te ontwikkel vir die massa-oordragseffektiwiteit van die plaat metaal pakking Flexipac 350Y deur gebruik te maak van verskillende eksperimentele- en modelleringstegnieke. Effektiwiteit is egter direk gekoppel aan hidroliese kapasiteit. Bestaande modelle in die literatuur is eers getoets om te bepaal of hulle die hidroliese kapasitiet van Flexipac 350Y akkuraat kan voorspel. Vir die doel is vloeistofterughou en drukval gemeet deur gebruik te maak van die sisteme lug/water en lug/swaar parafien. Alle eksperimente is in loodsaanlegskaal 200mm ID glaskolomme uitgevoer. Meeste van die modelle was relatief akkuraat in hulle berekening van die ladingspunt en die drukval vir die lug/water toets sisteem, maar was konsertief in voorspellings van die groothede vir die lug/swaar parafien sisteem. Effektiwiteit en drukval was gemeet deur gebruik te maak van die binêre toetssisteem chlorobenseen/etielbenseen onder totale terugvloei kondisies in 'n 200mm ID glaskolom. Daar is 'n groot verskil in die effektiwiteitsvoorspelling deur die verskillende modelle. Vervolgens is eksperimente ontwerp en uitgevoer om die dampfase massaoordragskoeffisiënt en die effektiewe oppervlakarea vir Flexipac 350Y onafhanklik te meet en te korreleer. Die dampfase massaoordragskoeffisient is gemeet en gekorreleer deur naftaleen te sublimeer vanaf spesiaal vervaardigde 350Y gestruktureerde pakking van metaalgaas. Die gebruik van numeriese vloeimeganika (CFD) om die dampfase massaoordragskoeffisient te bereken word gedemonstreer. Die effektiewe oppervlakarea vir dampfase massaoordrag is bepaal deur van 'n chemiese metode gebruik te maak. Die spesifieke absorpsietempo van CO2 in monoetanolamien (MEA) met n-propanol as oplosmiddel is gemeet in a benatte wand kolom en gebruik om die effektiewe oppervlakarea van Flexipac 350Y te bepaal op loodsaanlegskaal (200mm ID). Die effektiwiteit van Flexipac 350Y kon met 'n akkuraatheid van binne 9%gemodelleer word deur vloeistoffaseweerstand te ignoreer en van die korrelasies gebruik te maak wat in hierdie studie ontwikkel is. Die effektiwiteit en kapasiteit van die nuwe generasie hoë kapasiteit pakking Flexipac 350Y HC is ook gemeet en vergelyk met die normale kapasiteit pakking Flexipac 350Y. 'n Verhoging in kapsiteit van 20% is gemeet vir die HC pakking in vergelyking met die normale kapasiteit pakking vir die lug/water sisteem en 'n 4% verhoging in kapasiteit vir die lug/swaar parafien sisteem. Die verhoging in kapasiteit het gevarieër tussen 8% en 14% in die binêre totale terugvloei distillasie toetse en was afhanklik van die kolom druk. Die verhoging in kapasiteit was ten koste van 'n verlaging in effektiwiteit van ongeveer 3% onderkant die ladingspunt.

Thesis (PhDEng)--University of Stellenbosch, 2001. / ENGLISH ABSTRACT: Separation processes are an integral part of chemical engineering. The purity of a chemical product is among the principal factors influencing its value. Therefore, any method that can increase the purity of a product or decrease the cost of purification will have a direct effect on the profitability of the entire plant. An important class of separation processes is the solvent-based separations. This includes processes like extractive distillation, liquid-liquid extraction and chromatographic separation. Heterogeneous azeotropic distillation is closely related to these processes. The most important variable in the design of a solvent-based separation process is the choice of solvent. A genetic algorithm for the computer-aided molecular design of solvents for extractive distillation had been previously developed by the author. This algorithm was improved and expanded to include liquid-liquid extraction, heterogeneous azeotropic distillation, gas-liquid chromatography and liquid-liquid (partition) chromatography. At the same time the efficiency of the algorithm was improved, resulting in a speed increase of up to 500% in certain cases. An automatic parameter tuning algorithm was also implemented to ensure maximum efficiency of the underlying genetic algorithm. In order to find suitable entrainers for heterogeneous azeotropic distillation a method is required to locate any ternary heterogeneous azeotropes present in a system. A number of methods proposed in the literature were evaluated and found to be computationally inefficient. Two new methods were therefore developed for ternary systems. A methodology for applying these methods to quaternary and higher systems was also proposed. Two algorithms to design blended solvents were also developed. Blended solvents allow the use of simpler and thus cheaper solvents by spreading the active functional groups over several molecular backbones. It was observed in a number of cases that the blended solvents performed better than their individual components. This was attributed to synergistic interactions between these components. Experimental evidence for this effect was also found. The algorithm was applied to a number of industrially important separation problems, including the extremely difficult final purification process of alpha olefins. In each case solvents were found that are predicted to perform substantially better than those that are currently used in industry. A number of these predictions were tested by experiment and found to hold true. / AFRIKAANSE OPSOMMING: Skeidingsprosesse is 'n integrale deel van chemiese ingenieurswese. Die suiwerheid van 'n chemiese produk is een van die hoof faktore wat die waarde daarvan bepaal. Derhalwe sal enige metode wat die suiwerheid van 'n produk kan verbeter, of die koste van die suiwering daarvan kan verlaag, 'n direkte effek op die winsgewendheid van die hele aanleg hê. 'n Belangrike groep skeidingsprosesse is die oplosmiddel-gebaseerde skeidings. Dit sluit prosesse soos ekstraktiewe distillasie, vloeistofvloeistof ekstraksie en chromatografiese skeidings in. Heterogene azeotrope distillasie is nou verwant aan hierdie prosesse. Die belangrikste veranderlike in die ontwerp van so 'n oplosmiddel-gebaseerde proses is die keuse van oplosmiddel. 'n Genetiese algoritme vir die rekenaargesteunde molekulêe ontwerp van oplosmiddels vir ekstraktiewe distillasie is voorheen ontwikkel deur die skrywer. Hierdie algoritme is verbeter en uitgebrei om vloeistofvloeistofekstraksie, heterogene azeotrope distillasie, gas-vloeistof chromatografie en vloeistof-vloeistof (verdelings) chromatografie in te sluit. Ter selfde tyd is die doeltreffendheid van die algoritme verbeter, wat 'n verbetering in spoed van tot 500% in sekere gevalle tot gevolg gehad het. 'n Algoritme om die parameters van die onderliggende genetiese algoritme outomaties te verfyn is ook geïm plementeer om die optimale werksverrigting van die algoritme te verseker. Om gepaste saamsleepmiddels vir heterogene azeotrope distillasie te vind, word 'n metode benodig om enige ternêre heterogene azeotrope aanwesig in 'n stelsel op te spoor. 'n Aantal sulke metodes wat in die literatuur voorgestel is, is geëvaluEer en daar is gevind dat hierdie metodes ondoeltreffend is. Twee nuwe metodes is derhalwe ontwikkel vir ternêre stelsels. 'n Metodiek om hierdie metodes op kwaternêre en hoër stelsels toe te pas, is ook voorgestel. Twee algoritmes vir die ontwerp van gemengde oplosmiddels is ook ontwikkel. Gemengde oplosmiddels laat die gebruik van eenvoudiger en dus goedkoper oplosmiddels toe, deur die aktiewe funksionele groepe oor 'n aantal molekulêe strukture te versprei. Daar is 'n aantal gevalle waargeneem waar die mengsel beter skeiding bewerkstellig het as die individuele oplosmiddels waaruit dit bestaan. Dit is toegeskryf aan 'n sinergistiese wisselwerking tussen die komponente van die mengsel. Eksperimentele getuienis vir hierdie effek is ook ingewin. Die algoritme is toegepas op 'n aantal belangrike skeidingsprobleme vanuit die bedryf, insluitende die uiters moeilike finale suiwering van alfa olefiene. In elke geval is oplosmiddels gevind wat volgens voorspelling aansienlike beter skeidings sal bewerkstellig as dié wat tans in die bedryf gebruik word. 'n Aantal van hierdie voorspellings is eksperimenteel getoets en korrek bewys.